Finding location of note points for notes in a note image

Question

I want to start a project that uses a very basic form of recognition of optical music.

For those who understand the notes:. Unlike other OMR projects, the only information to be extracted is the order and height of each note in the panel. Square notes, half notes and whole notes need to be distinguished. Shorter notes can be thought of as quarter notes. Dots on notes can be ignored. Speaker marking is not important

For everyone: Strictly speaking, I need to find the places of each of the following ...

... in a pattern like this ...

I have no experience with image processing, so it would be very helpful to understand a basic conceptual explanation of which method or set of methods is used to achieve this.

Answer 1

I would do the following:

Extract lines using the Hough transform. (You also get a corner). Trim each group of lines (5 lines) and process individually.

For each group of lines, you know the angle of the lines, so you can get the location of the vertical small lines that separate the stripes. Repeat the search in the Hough space, but with a certain angle. (Original + 90). Trim each bar and process individually.

For each bar, use pattern matching on possible notes (Quarter, Half, etc.)

Answer 2

I did something similar to your work and trusted me that it was a complete mess.

However, for the pitch for each note, you take the head out of the rest and calculate the barycenter and compare its position with the position of the lines calculated with the Hough transform, as already mentioned (assuming that the lines are already straight: if not I think you can use Fourier transform).

All the time you need a classification algorithm.